Implicit conversion from float -15.477707006369428 to int loses precision